Pendlebury ST, Chen PJ, Bull L, Silver L, Mehta Z, and Rothwell PM, for the Oxford Vascular Study. Methodological Factors in Determining Rates of Dementia in Transient Ischemic Attack and Stroke: (I) Impact of Baseline Selection Bias. Stroke. 2015
Rates of dementia in the first year after stroke have varied in prior studies. In this report, Pendlebury et al evaluated the methods used for measuring TIA/stroke-related dementia rates to better understand potential sources of bias. They looked at the impact of various study entry criteria on the pre- and post-event rate of dementia from a large population-based study.
Patients with TIA or stroke (as defined by the WHO criteria) from 2002-2007 were assessed from the Oxford Vascular Study, a prospective cohort study from Oxfordshire, UK. Cognitive testing was done using the MMSE, MoCA (or telephone MoCA), and telephone interview for cognitive status (TICS) assessment at 1, 3, and 6 months as well as 1, 5, and 10 year follow-up. Pre-cerebrovascular event dementia was recorded when it was a listed diagnosis at the time of the index TIA/stroke (from documentation by primary care provider) or determined by physician review of prior patient records. Post-event dementia was diagnosed when MMSE<24 or MOCA<20, telephone MOCA<9, or TICS<22 during follow-up assessment (those with pre-event dementia were excluded). For those who could not complete testing (due to aphasia, visual field loss, etc.), primary care physician records and collateral history from patient informant/family was used to determine cognition using the DSM-IV criteria.
1,234 patients were evaluated – 33% with TIA, 37% with minor stroke (NIHSS≤3), 30% major stroke, and 5% ICH. 1189 patients were alive at ascertainment and 1097 patients completed baseline assessment. The authors found that pre-stroke dementia rates were >2 times higher in those who died early. This resulted in a true population-based pre-event dementia rate that was one-third higher than when measured in the 1097 surviving patients. When excluding patients who were age>80 years or functionally impaired (mRS≥3), the pre-event dementia rate was cut in half – from 8% to 3% (p<0.001). Excluding aphasic patients or those with significant co-morbidities reduced the pre-event dementia rate by about a quarter. Rates were ~2-fold higher in hospitalized compared to non-hospitalized patients in both pre-event (10% vs 6%, p=0.01) and post-event (24% vs 11%, p<0.0001) dementia. Newly diagnosed post-event dementia was three times more common in those >80 years old (27% vs 9%, p<0.0001) and those with prior functional impairment (39% vs 10%, p<0.0001); it was also significantly higher in aphasic patients (22% vs 15%, p=0.02).
This study demonstrates that various selection criteria can influence pre- and post-event dementia rates when applied to a large cohort. Prior studies have excluded patients who were older, aphasic, had baseline functional impairment, and those who could not be available for study assessment. As shown in this report, this can have a substantial effect on measured rates of dementia after stroke/TIA.
The authors used collateral information – from primary care provider records, patient informants/family, etc. – to determine cognition in those who could not be formally tested in this study. They also depended on cognitive assessment scores for determining dementia rather than a making a true clinical diagnosis. Despite the limitations of the manuscript, a valid point is made – future studies that assess pre- and post-stroke/TIA cognitive impairment should be as inclusive as possible to maximize validity.